performance evaluation of an adaptive ofdm- based plc

10467

ISSN 2286-4822

www.euacademic.org

EUROPEAN ACADEMIC RESEARCH

Vol. II, Issue 8/ November 2014

Impact Factor: 3.1 (UIF)

DRJI Value: 5.9 (B+)

Performance Evaluation of an adaptive OFDM-

based PLC System with an Impulsive Noise

Environment

ALAA A. GHAITH Department of Physics & Electronics

Faculty of Sciences, Lebanese University

Lebanon

Abstract:

Power Line Communication (PLC) is an interesting approach

in establishing last mile broadband access in rural areas. PLC

provides a ready medium for broadband internet connectivity and as

well as monitoring and control functions for both industries and

homes. In fact, PLC network is the most expansive network in the

world reaching every room. However, it provides an available channel

that is hostile when used for communication purposes. This hostility is

due to the many problematic characteristics of the PLC from a

communications perspective. They include reflections due to impedance

mismatch, multipath due to the cable joints, as well as the different

types of noise inherent in the channel. In this paper, we are interested

by the impact of impulse noise on the performance of an adaptive

OFDM based PLC system. The article presents a design of the PLC

model, which is composed of communication model, model of power

line and noise model. The communication model is realized as an

adaptive OFDM system, power-lines are modelled like a multipath

channel. Noise model is modelled as white noise in addition to the

impulsive noise, the mathematics behind the impulse is utilized in an

attempt to fully characterize the impulsive noise. The impedance

mismatching is considered and, with the channel model, are estimated

at the receiver. On the resulting PLC communication model was shown

comparison of different modulation technique with the adaptive

system. Different schemes were simulated to compare the bit error rate

(BER) for different impulsive noise parameters.

Alaa A. Ghaith- Performance Evaluation of an adaptive OFDM-based PLC

System with an Impulsive Noise Environment

EUROPEAN ACADEMIC RESEARCH - Vol. II, Issue 8 / November 2014

10468

Key words: power line, OFDM, impulsive noise, impedance

mismatch, modelling, modulation, and channel estimation.

1. Introduction

Most of us today have a wireless network in our homes to

enable us to enjoy unfettered access to the internet and share

data between PCs and networked peripherals like printers.

More recently the trend towards connecting into home

entertainment devices using ethernet ports to connect to TVs,

Blu-ray players, and gaming devices has become increasingly

common place. Kaspersky Labs estimates that the current

number of UK homes with a wireless LAN installed is around

57 per cent. Ofcom estimates that 1.5 million households (out of

a total of approximately 22 million households) have deployed

PowerLine Technology to connect them up. The most

interesting aspect regarding PLC is the ability to use existing

infrastructure for signal transmission, reducing deployment

costs. Until recently the biggest problem for PLC was the low

distance coverage and low data rate which was eliminated with

the introduction of new standards.

The PLC is a need to use distribution lines of electricity

for the control signals, IP telephony transfer, and remote data

acquisition [Mlynek 2008], [Achmad 2007], and [Koinakis

2009]. It comes up almost simultaneously with electrical power

network developing. This technology becomes more and more

important. The PLC technology should be as an alternative to

other existing data channels [Orgoň 2007].

PLC technologies are mainly classified into two;

narrowband and broadband depending on the frequency band of

operation. Broadband PLC operates in the frequency range

between 1MHz to 300 MHz. Narrowband PLC operates in the

frequency range between 3 kHz to 500 kHz. PLC networks also

form part of local area networking solutions [Berger 2013]

[Ferreira 1996]. However, the PLC channel is a hostile

environment for use as a communications media. This is




10469

primarily so because, the channel characteristics are highly

varying with frequency, time, loads and topology. The channel

is also plagued with different sources of noise which are

difficult to effectively describe parametrically. The main noise

types in PLC include background noise, narrowband

interference and impulse noise. Also the signal is attenuated as

it traverses the channel from the transmitter to the receiver.

This attenuation is mainly dominated by frequency selective

fading. The proper choice of modulation techniques, channel

estimation techniques, and impedance mismatching correction

methods for the PLC channel is a thorny issue [Berger 2013],

[Ferreira 1996] [Zwane 2014], [Lazaropoulos 2012],

[Zimmermann 2002], [Mulangu 2011] and [Mulangu 2012]. The

biggest threat to PLC performance is noise; and impulsive noise

is the most dominant. This noise and its impact on the

performance of an adaptive OFDM based system is the main

focus of this paper.

2. Impulsive Noise

Impulsive noise is the most severe form of noise in PLC

channels. It is mainly caused by the switching ON/OFF of

electrical appliances or faults in the network. It is classified

into three main categories [Zimmermann 2002]:

1- Impulsive noise that is synchronous with the mains

frequency and is periodic: This type of impulsive noise is

cyclostationary and is synchronous with the mains

frequency. It is generated by silicon controlled rectifiers in

different power supplies.

2- Impulsive noise that is periodic but asynchronous with the

mains frequency: This category of impulsive noise is

generated by periodic impulses whose repetition rates are

between 50 to 200 kHz.

3- Asynchronous impulsive noise: This kind of impulsive noise

is very unpredictable and is the most dominant. It exhibits




10470

no regular occurrence and mainly arises from transients

that originate from the connection or disconnection of

electrical appliances from the power line network. The

impulses can last for between some microseconds and a few

milliseconds with a random occurrence. The power spectral

density for this kind of noise can go as high as 50 dB above

the background noise. It is also sometimes referred to as

sporadic impulsive noise.

The four basic parameters that describe impulsive noise are the

impulse width tw, arrival time tarr, inter-arrival time tIAT or the

impulse distance td and the impulse amplitude, A. The impulse

width is the time duration that an impulse event lasts. The

inter-arrival time is defined as the time difference between the

start of two consecutive impulse events. The impulse distance is

the time between the end of an impulsive event and the

beginning of another. Thus, it defines the frequency of

occurrence of the impulsive noise. The three basic impulsive

event time parameters are related by the following expression:

, 1 ,IAT w d arr i arr it t t t t (1)

And, using a general impulse function imp(t) with unit

amplitude and unit width, a train of impulses can be described

by [Zimmermann 2000]:

,

1 ,

.N

arr i

imptrain i

t w i

t tn t A imp

t

(2)

where nimptrain(t) is the train of impulses. The parameters A, tw

and tarr, can be used to derive secondary parameters that are

crucial in analysis of impulse noise and studying its

behavioural characteristics over time. One of these secondary

parameters is the impulse rate, given by [Zimmermann 2002]:




10471

imp

imp

win

Nr

T

(3)

where Nimp is the number of impulses that occur within a given

window of observation Twin. Another key aspect would be the

actually disturbed time, which can be determined from the

“disturbance ratio”; which by definition is the ratio of the sum

of the widths of all impulses generated within a window of

observation, and the length of the window, that is

[Zimmermann 2002]:

,1

impN

w ii

win

tdisturbance ratio

T

(4)

3. System Model

For a creating of the complete PLC communication system,

there is necessary to create model of channels as well as noise

model and a transmitter and a receiver models. The complete

PLC model will be created from particular models. There will be

possible to create analysis of a concrete power line based on the

simulations of this system with various models of lines. The

analysis will be possible to judge in term of possibility to using

of various combinations of PLC technologies, security transfer,

modulations etc. So that there will be obtained to best

parameters of data transfer in mentioned systems. It is

necessary to create the channel models for the PLC simulation.

There are more possibilities of power line model creating. First

of them is the power line model as environment with multipath

signal propagation. The parameters of this line are obtained

from a distribution network topology or based on metering.

Second of them is model, which applies chain parameter

matrices to describing the relation between input and output

voltage and current by two-port network.




10472

In design of PLC system, it is necessary to bear in mind the

character of transmission medium and interferences which the

PLC communication is influenced. It is necessary to find

useable guard interval, modulation technique and encryption to

ensure of security and fool proof communication with smallest

error rate between data source and receiver. The PLC

communication system is possible to divide to particular parts

for purpose of modelling:

1- PLC communication model

2- Models of power lines,

3- Noise model.

3.1. The PLC communication model: Adaptive OFDM

The model of PLC communication is created by a transmitter,

receiver and channel block. It serves for a creating of a source

and destination of data communication for subsequent

simulations of lines model which they are replaced by block of

channel. The basic PLC communication model with OFDM

system is shown in the Fig.1

Figure 1: The Adaptive OFDM based PLC system.

The control block is used only for enabling or disabling some

functionalities of the system, like the estimation/equalization

schemes, the start block is responsible of generating the

random stream of bits and selecting the corresponding mapping

in each sub-carrier with respect to the SNRs estimated by the

channel estimation block. No error correcting code is used




10473

which in addition to an interleaver can improve the

performances presented below. A stream of symbols is resulting

from the mapping; this serial data obtained is converted to

parallel and the pilots are added which are necessary to include

to the transmission in case of continuous channel estimation.

The estimation is important for determination of amplitude and

phase of map’s constellation each of subcarrier. The estimation

of the channel in an OFDM system requests a inserting of

known symbols or a pilot structure to the OFDM signal. The

number of parallel streams resulting from the data and pilots

should match to the number of carriers. IFFT block transforms

data from frequency to time domain. A protect interval is used

in OFDM to prevent of ISI (inter symbol interference). A cyclic

prefix (CP) is created by a few of last samples of OFDM

symbols. CP creates a protect interval between adjacent

transferred OFDM symbols in time area. This is a way how to

keep orthogonally carries. Again, the parallel streams are

converted into serial, and an interpolation filter is used to

convert the digital signal into analog to be upconverted by the

modulator to a certain carrier frequency, here, fc = 46.5 MHz.

Figure 2: The transmitted signal

Figure 2 shows the power spectral density of the transmitted

signal. The variation in the power level presented in Fig.2 is




10474

due to the different level of mapping, so different power level,

used in the transmitted signal.

After the channel which is considered as a multipath

channel in addition to the noise sources added, the receiver’s

blocks will do the inverse work of the transmitter’s blocks. So

the demodulator will down-convert the signal into baseband,

and the decimation filter is used to digitize the signal. The

cyclic prefix is removed and the signal is transformed into

frequency domain. The most important blocks here are: the line

impedance estimation, the channel estimation, and the

equalization & detection blocks.

The line impedance estimation block works only in the

first channel use, where all the data sent from the transmitter

are known at the receiver and used for this estimation. This

block is represented in Fig. 3, it is very simple where we should

make a certain division with the known data (Training) and

after that an average is done to obtain the impedance estimated

on each sub-carrier. These impedances are forward to the

channel estimation block and the equalization & detection block

for the next channel use.

Figure 3: The block diagram of impedance estimation.

For the second channel use, the impedance estimator will be

disabled and the channel estimator block and the equalizer

block can work normally starting from this second channel use

until the end of the connection. In this time, the signal contains

information data and a small training sequence used by the

channel estimator represented in Fig. 4.

The channel estimation block is responsible of

estimating the channel taps and estimating the SNR on each




10475

sub-carrier. The estimation of the channel taps is done simply

by doing a division between the known training sequence and

the corresponding received sequence, since we work here in the

frequency domain. For the SNRs estimation, we are obligate to

work in each sub-carrier trying to separate the noise from the

desired signal, and compute the power of desired signal divided

by the power of the estimated noise. When we obtain these

SNRs, they will be sent back to the transmitter and saved at

the receiver for the next use, which choose the mapping level on

each subcarrier in corresponding to the SNRs received and a

certain threshold chosen by the method represented by S(i) in

Fig. 5 and memorized in the start block at the transmitter.

Figure 4: The block diagram of channel estimation

Since we work in the frequency domain here after the FFT, so

the equalizer’s work is very simple, it has just to divide the

received sequence by the estimated channel (make sure that

the estimated channel is in frequency domain). The resulting

sequence is than demapping with a different de-mapper on each

sub-carrier. Finally the bit stream outputted by the de-mapping

is compared with the random bits generated at the transmitter

and the BER is calculated.

The transmission rate (or spectral efficiency) of each

transmission is basically determined by the modulation and

channel coding combination. Three main modulation schemes

are considered here: i) Quadrature Phase Shift Keying (QPSK),

ii) 16-Quadrature Amplitude Modulation (16-QAM) and iii) 64-




10476

QAM. In each case, the modulation efficiency is equal to the

number of bits per symbol that can be sent. If the channel

coding is used the efficiency is multiplied by the coding rate.

According to a fixed Target BER, which accounts for application

or service types, the minimum SNR threshold is set to achieve

the desired mapping scheme as shown in the Fig. 5. We note

here that the target BER chosen for those values are assumed

to be equal to 10-3, but in general it may to take a positive value

depending on the service offered.

Figure 5: SNR thresholds selection for different mapping

3.2. PLC Channel and Additive Noise Model

Besides simulation of communication system characteristic, it

is necessary to identify possible sources of interference and

noise because the power line has a significant attenuation of

signal and various noises. Therefore the data transfer has a

high error rate without any checking algorithm. The

fundamental influence on data transmission over power lines

are mainly the negative characteristics of power networks.

These characteristics can be summarized in:

Figure 6: PLC channel model.




10477

1- Mismatched impedance

2- Attenuation on the communication channel

3- Noise and Noise changing in time.

Figure 6 shows a simplified block model of the PLC channel, in

which the described characteristics and parameters are

included. The parameters of interference, except noise, are

represented as a time variable linear filter described by the

frequency parameters. Noise is depicted as additive random

process. This model captures the whole range of parameters

which are necessary for a model of the communication system

with corresponding characteristics, although this model is

schematically simplified in the figure. The impulse response of

the linear filter and the noise can be either estimated from the

measurement or derived from the theoretical analysis. Here,

they are computed by a huge number of measurements and the

average of all taken measurements is considered as a

representative model for the linear filter and the summation of

different type of noise presented in Fig. 6. The impulse response

and the frequency response of the linear filter is represented in

Fig.7 and figure 8 shows an one shot time domain value of the

noise considered here.

Figure 7: The time variable linear filter model used




10478

Figure 8: One shot time domain representation of noise

4. Simulations Results and Analysis

A key performance index to evaluate the performance is the

BER given a received SNR over the considered channel

presented above. We consider a received SNR from 0 up to 100

dB and examine the BER. The received signal to noise ratio is

considered here as Eb/N0 where Eb is the received energy per

bit, and N0 is the total noise power spectral density. Monte

Carlo simulation by MatLab is used to obtain the results shown

in the following figures; in despite that the block diagrams

shown before were done using Simulink. For the evaluation,

three mapping levels, QPSK, 16QAM, and 64QAM are

considered and finally, compared with the adaptive based

OFDM system over a PLC channel.

4.1. Comparison of different mapping level

The testing of different types of modulations for data

communication over power line has been accomplished on

created model. Figure 9 shows a comparison of the QPSK,

16QAM, 64QAM, and 256QAM modulations in OFDM system.

Here, the amplitude of the impulsive noise is considered equal

to 1 (A=1); in despite that it can be much smaller than this

value. The idea behind this assumption was to compare the

effect of the modulation alone.




10479

Figure 9: Performance of different mapping levels (A=1)

The results in Fig. 9 show that the performance of all mapping

levels are very bad, this because of our assumption, in despite

that the performance of QPSK is much better than the others,

and the performances decrease when the mapping level

increase. This conclusion can be done easily when we compare

these mapping levels over AWGN channel, so it was expected.

But the error floor beginning from 30 dB for each curve was not

expected and it can be justified by the presence of the impulsive

noise all time and whatever the SNR is, so when the SNR is

large the power of AWGN noise is small but the power of the

impulsive noise remains fixed, which mean that this error floor

is the best performance in presence of this type of noise.

However, these performances can be improved using an error

correction code or using one of the criteria of adaptation when

we fixed a certain threshold for the SNR per sub-carrier and if

it is smaller than this threshold we don’t transmit anything

over this sub-carrier. Thus the total data rate of the system is

decreased but we obtain a good improvement of the

performance shown in Fig. 10 and 11. We remark in Fig. 10

that the performance of QPSK is much more improved than the

other mapping levels, which indicates that working with QPSK

is more preferable for all applications with these channel

conditions, in despite that it gives the smallest data rate. In




10480

figure 11, we can show that at high SNRs the total data rate of

the system using the threshold elimination (called with

elimination; some sub-carriers are not used for transmission, so

eliminated) remains lower than the data rate of the system

without any sub-carrier elimination (called complete; all sub-

carriers are used for transmission) which is fixed to the

maximum possible with each mapping level with respect to the

bandwidth and this loss is approximately between 26% and

28%, and we can remark also that this mapping level cannot

respect the conditions of the threshold so they cannot send any

data, thus in this interval of SNR we should use the BPSK

modulation, whatever the loss obtained in the data rate.

Figure 10: Performance of mappings (A=1) with Elimination

Figure 11: Data Rate of mappings (A=1) with Elimination




10481

4.2. The effect of the impulsive noise amplitude

In this simulation, we want to study the effect of the impulsive

noise amplitude, and the results shown in Fig. 12 and 13

confirm that the error floor presents in the curves is decreasing

with the impulsive noise amplitude. We obtained a

representative value of this amplitude during the

measurements done for deducing the model of the PLC channel,

and this value was approximately equal to 0.0316 (A2 = 0.001).

Figures 12 and 13 show the performance of considered mapping

levels for both amplitudes A=1 and A=0.0316. Figure 12

represents the system without elimination and figure 13

represents the system with elimination.

Figure 12: Performance of mappings (A=0.0316) without Elimination

Figure 13: Performance of mappings (A=0.0316) with Elimination




10482

Figure 12 shows that the performance of all mapping levels is

improved, in despite that, the BER remains high, the curves

are very close to each other, especially at high SNR. From Fig.

13, we can conclude two things; the first conclusion is that, with

A=0.0316, the performances are improved for all mappings

except the QPSK mapping where we remark that the curve

converges slower to the error floor value which is smaller than

the one with A=1; the second conclusion is common with Fig. 12

where the performances are approaching to each other.

4.3. Comparison with the adaptive OFDM system

Here, we will present the advantage adaptive system. The idea

behind this work is to consider that we have a certain

application which needs a minimum BER to work properly, so

we consider this minimum BER as a target BER and we

compute the threshold corresponding to the mappings used.

When the SNR in a certain sub-carrier is bigger than or equal

the threshold of 16QAM, for example, we can transmit over this

sub-carrier this modulation, otherwise the mapping with lower

level is used, and so on each sub-carrier and for every channel

use. Here, the target BER is fixed to 10-3 and when this BER

cannot be obtained the lowest mapping level is used. The PLC

channel mode chosen is the more realistic one which

corresponds to any practical use of a PLC system, it considers

the presence of impulsive noise with amplitude A=0.0316.

Figures 14 and 15 show the evolution of the BER and the data

rate when the SNR is increasing, and to be fair these adaptive

system performances are compared with the system with

elimination. We remark clearly that, after 37 dB, the target

performance is obtained, so the adaptive system will try to

increase the data rate by carrying upper mapping level on the

sub-carriers without affecting the BER; for these reasons, we

obtained in Fig. 15 a data rate higher than the one of 64QAM

and in the same time the BER performance of our system

remains confirmed to the desired application, in despite that




10483

the BER of 64QAM is largely higher than the target one. As

shown in Fig. 15, the data rate obtained is increasing with the

SNR until a certain value which is nearly 390 Mbps for an

acceptable probability of error.

Figure 14: Performance of Adaptive Modulation (A=0.0316)

Figure 15: Data Rate of Adaptive Modulation (A=0.0316)

5. Conclusion

The progress in PLC technology has come about in the last

decade. PLC technology is seemed as an alternative data

channel. The article deals with design of the PLC

communication system model. The model is composed of the

OFDM communication model, the model of power lines and

noise model. In this paper, an adaptive approach has been




10484

presented that assesses the impact of impulsive noise on an

adaptive OFDM-based PLC system. The statistical properties of

the impulsive noise have been utilized in their most basic form.

The error statistics obtained show that the performance of the

system is highly limited in the presence of the impulsive noise

in the channel. We observe that a lower value of the amplitude

translates to more frequent impulses and this improves the

error performance of the system. Thus to improve on the system

performance, an error correction code can be used, it is our case

here, so the adaptive system should be implemented on the

channel. We show two level of adaptive system; the first one is

to do not use some sub-carriers for transmission if their SNR is

below a certain threshold without changing the mapping level.

The second level is to use this threshold to adapt the mapping

level of the modulation with respect to the target performance

desired. The results presented here show that we can obtain a

good improvement with this technique. From the results of

simulations it follows that an inappropriate choice of

modulation can significantly affect the resulting signal. The

error correction code is not used here. In the future work, we

will design a corresponding channel code which will be adaptive

and specific to this application. After that, the complex adaptive

OFDM-based PLC system can be used for future

standardization. The results of simulations based on the model

will be compared with measurements in the future work.

REFERENCES

[Mlynek 2008] P. Mlynek, J. Misurec, M. Koutny, ”The

communication unit for remote data acquisition via the

Internet,” In Proceedings of the 7th WSEAS

International Conference on Circuits, systems,

electronics, control and signal processing (CSES'08).

Puerto de La Cruz, Spain: WSEAS Press, 2008.




10485

[Achmad 2007] R. Achmad, U. Agung, F. Tetsuo, K. Hidehiro,

Y. T. Kyaw, U. Yoshiyori, U. Yuki, U. Yosuke , “Wireless

LAN and power line communication platform for e-

learning multimedia system in underdeveloped area in

Lombok island,” in 6th WSEAS Int. Conf. on Electronics,

Hardware, Wireless and Optical Communications, pages

23–28, 2007.

[Koinakis 2009] C. J. Koinakis, J. K. Sakellaris, “Integration of

building envelope and services via control technologies,”

in Proceedings of the 13th WSEAS international

Conference on Systems, Rodos, Greece, July 22 - 24,

2009.

[Orgon 2007] M. Orgon, “PLC/BPL and Next Generation

Networks,” in POWERCOM - Conference

Communication over MV and LV power lines, Praha,

2007.

[Berger 2013] L. T. Berger, A. Schwager, and J. J. Escudero-

Garzas, “Power line communications for smart grid

applications,” Journal of Electrical and Computer

Engineering, Vol. 2013, Article 712376, 16 pages, 2013.

[Ferreira 1996] H. C. Ferreira, et al., “Power line

communications: An overview,” IEEE AFRICON 1996,

558-563, Stellenbosch, September 24-27, 1996.

[Zwane 2014] F. Zwane, and T. J. Afullo, “An alternative

approach in power line communication channel

modelling,” Progress In Electromagnetics Research C,

Vol. 47, 85-93, 2014.

[Lazaropoulos 2012] A. G. Lazaropoulos, “Broadband

transmission characteristics of overhead high-voltage

power line communication channels,” Progress In

Electromagnetics Research B, Vol. 36, 373-398, 2012.

[Zimmermann 2002] M. Zimmermann, and K. Dostert, “A

multipath model for the power line channel,” IEEE

Transactions on Communications, Vol. 50, No. 4, 553-

559, April 2002.




10486

[Mulangu 2011] C. T. Mulangu, T. J. Afullo, and N. M. Ijumba,

“Modelling of broadband powerline communication

channels,” SAIEE, Vol. 102, No. 4, 107-112, December

2011.

[Mulangu 2012] C. T. Mulangu, T. J. Afullo, and N. M. Ijumba,

“Estimation of specific attenuation due to scattering

points for broadband PLC channels,” PIERS

Proceedings, Kuala Lumpur, Malaysia, 92-96, March 27-

30, 2012.

[Zimmermann 2002] M. Zimmermann, and K. Dostert,

“Analysis and modelling of impulsive noise in broad-

band powerline communications,” IEEE Transactions on

Electromagnetic Compatibility, Vol. 44, No. 1, 249-258,

November 2002.

[Zimmermann 2000] M. Zimmermann, and K. Dostert,

“Analysis of the broadband noise scenario in powerline

networks,” Proc. of ISPLC, 131-138, 2000.

performance evaluation of an adaptive ofdm- based plc

Documents